Search Results (546)

The V/G criterion is a critical indicator for monitoring dynamic changes during Czochralski silicon single crystal (Cz-SSC) growth. However, the inability to measure it in real time forces reliance on offline feedback for process regulation, leading to imprecise control and compromised crystal quality. To overcome this limitation, this paper proposes a novel soft sensor modeling framework that integrates both mechanism-based knowledge and data-driven learning for the real-time prediction of the crystal quality parameter, specifically the V/G value (the ratio of growth rate to axial temperature gradient). The proposed approach constructs a hybrid prediction model by combining a data-driven sub-model with a physics-informed mechanism sub-model. The data-driven component is developed using an attention-based dynamic stacked enhanced autoencoder (AD-SEAE) network, where the SEAE structure introduces layer-wise reconstruction operations to mitigate information loss during hierarchical feature extraction. Furthermore, an attention mechanism is incorporated to dynamically weigh historical and current samples, thereby enhancing the temporal representation of process dynamics. In addition, a robust ensemble approach is achieved by fusing the outputs of two subsidiary models using an adaptive weighting strategy based on prediction accuracy, thereby enabling more reliable V/G predictions under varying operational conditions. Experimental validation using actual industrial Cz-SSC production data demonstrates that the proposed method achieves high-prediction accuracy and effectively supports real-time process optimization and quality monitoring. Full article

Salient object detection in optical remote sensing images has attracted extensive research interest in recent years. However, CNN-based methods are generally limited by local receptive fields, while ViT-based methods suffer from common defects in noise suppression, channel selection, foreground-background distinction, and detail enhancement. To address these issues and integrate long-distance contextual dependencies, we introduce GDUFormer, an ORSI-SOD detection method based on the ViT backbone and Gated Differential Units (GDU). Specifically, the GDU consists of two key components—Full-Dimensional Gated Attention (FGA) and Hierarchical Differential Dynamic Convolution (HDDC). FGA consists of two branches aimed at filtering effective features from the information flow. The first branch focuses on aggregating spatial local information under multiple receptive fields and filters the local feature maps via a grouping mechanism. The second branch imitates the Vision Mamba to acquire high-level reasoning and abstraction capabilities, enabling weak channel filtering. HDDC primarily utilizes distance decay and hierarchical intensity difference capture mechanisms to generate dynamic kernel spatial weights, thereby facilitating the convolution kernel to fully mix long-range contextual dependencies. Among these, the intensity difference capture mechanism can adaptively divide hierarchies and allocate parameters according to kernel size, thus realizing varying levels of difference capture in the kernel space. Extensive quantitative and qualitative experiments demonstrate the effectiveness and rationality of GDUFormer and its internal components. Full article

Given the high cost of acquiring large-scale annotated datasets, few-shot object detection (FSOD) has emerged as an increasingly important research direction. However, existing FSOD methods face two critical challenges in remote sensing images (RSIs): (1) features of small targets within remote sensing images are incompletely represented due to extremely small-scale and cluttered backgrounds, which weakens discriminability and leads to significant detection degradation; (2) unified classification boundaries fail to handle the distinct confidence distributions between well-sampled base classes and sparsely sampled novel classes, leading to ineffective knowledge transfer. To address these issues, we propose TS-FSOD, a Transfer-Stable FSOD framework with two key innovations. First, the proposed detector integrates a Feature Enhancement Module (FEM) leveraging hierarchical attention mechanisms to alleviate small target feature attenuation, and an Adaptive Fusion Unit (AFU) utilizing spatial-channel selection to strengthen target feature representations while mitigating background interference. Second, Dynamic Temperature-scaling Learnable Classifier (DTLC) employs separate learnable temperature parameters for base and novel classes, combined with difficulty-aware weighting and dynamic adjustment, to adaptively calibrate decision boundaries for stable knowledge transfer. Experiments on DIOR and NWPU VHR-10 datasets show that TS-FSOD achieves competitive or superior performance compared to state-of-the-art methods, with improvements up to 4.30% mAP, particularly excelling in 3-shot and 5-shot scenarios. Full article

Semantic segmentation of high-resolution remote sensing imagery is a critical technology for the intelligent interpretation of sensor data, supporting automated environmental monitoring and urban sensing systems. However, processing data from dense urban scenarios remains challenging due to sensor signal occlusions (e.g., shadows) and the complexity of parsing multi-scale targets from optical sensors. Existing approaches often exhibit a trade-off between the accuracy of global semantic modeling and the precision of complex boundary recognition. While the Segment Anything Model (SAM) offers powerful zero-shot structural priors, its direct application to remote sensing is hindered by domain gaps and the lack of inherent semantic categorization. To address these limitations, we propose a dual-branch cooperative network, PriorSAM-DBNet. The main branch employs a Densely Connected Swin (DC-Swin) Transformer to capture cross-scale global features via a hierarchical shifted window attention mechanism. The auxiliary branch leverages SAM’s zero-shot capability to exploit structural universality, generating object-boundary masks as robust signal priors while bypassing semantic domain shifts. Crucially, we introduce a parameter-efficient Scaled Subsampling Projection (SSP) module that employs a weight-sharing mechanism to align cross-modal features, freezing the massive SAM backbone to ensure computational viability for practical sensor applications. Furthermore, a novel Attentive Cross-Modal Fusion (ACMF) module is designed to dynamically resolve semantic ambiguities by calibrating the global context with local structural priors. Extensive experiments on the ISPRS Vaihingen, Potsdam, and LoveDA-Urban datasets demonstrate that PriorSAM-DBNet outperforms state-of-the-art approaches. By fine-tuning only 0.91 million parameters in the auxiliary branch, our method achieves mIoU scores of 82.50%, 85.59%, and 53.36%, respectively. The proposed framework offers a scalable, high-precision solution for remote sensing semantic segmentation, particularly effective for disaster emergency response where rapid feature recognition from sensor streams is paramount. Full article

Hyperspectral image (HSI) classification is pivotal in remote sensing, yet deep learning models, particularly Transformers, remain susceptible to spurious spectral–spatial correlations and suffer from limited interpretability. These issues stem from their inability to model the underlying causal structure in high-dimensional data. This paper introduces the Causal Attention Transformer (CAT), a novel architecture that integrates causal inference with a hierarchical CNN-Transformer backbone to address these limitations. CAT incorporates three key modules: (1) a Causal Attention Mechanism that enforces temporal and spatial causality via triangular masking and axial decomposition to eliminate spurious dependencies; (2) a Dual-Path Hierarchical Fusion module that adaptively integrates spectral and spatial causal features using learnable gating; and (3) a Linearized Causal Attention module that reduces the computational complexity from $O\left(N^2\right)$ to $O\left(N\right)$ via kernelized cumulative summation, enabling scalable high-resolution HSI processing. Extensive experiments on three benchmark datasets (Indian Pines, Pavia University, Houston2013) demonstrate that CAT achieves state-of-the-art performance, outperforming leading CNN and Transformer models in both accuracy and robustness. Furthermore, CAT provides inherently interpretable spectral–spatial causal maps, offering valuable insights for reliable remote sensing analysis. Full article

In the field of energy transformation, the application of batteries is widening. To address the challenge of health state estimation of energy storage batteries with multiple operating conditions, this study analyzes the aging cycle operation data of lithium-ion batteries and develops a scheme to extract a number of raw features and their corresponding health status labels. Multidimensional candidate feature sets that capture aging information under different conditions are constructed. Subsequently, a three-stage feature selection strategy, including Pearson and Spearman correlation analysis, hierarchical redundancy elimination, and minimum redundancy maximum relevance, was applied to screen the candidate feature set of each condition, resulting in customized feature sets with condition adaptability. By analyzing the occurrence frequency and mean absolute correlation coefficient of each feature within the custom feature set, a comprehensive feature set with multi-condition adaptability was screened and determined. On this basis, by integrating temporal sequence information and operating condition information, a dual-path fusion estimation model with attention mechanism and condition modulation was established. The validation results of the lithium-ion battery multi-condition cycling aging dataset demonstrate that the model achieves accurate health state estimation, with mean absolute error and root mean square error of 0.8281% and 0.9835%, respectively. Finally, comparisons with other methods were conducted in terms of feature selection strategies and model estimation performance. The results demonstrate that the proposed approach achieves superior estimation accuracy and enhanced interpretability. Full article

In order to solve the problems of small-target detection in UAV aerial photography, such as small scale, blurred features and complex background interference, this article proposes the ACS-YOLOv8s method to optimize the YOLOv8s network: notably, most small man-made targets in UAV aerial scenes (e.g., small vehicles, micro-drones) inherently possess symmetry, a key geometric attribute that can significantly enhance the discriminability of blurred or incomplete target features, and thus symmetry-aware mechanisms are integrated into the aforementioned improved modules to further boost detection performance. The backbone network introduces an adaptive feature enhancement module, the edge and detail representation of small targets is enhanced by dynamically modulating the receptive field with deformable attention while also capturing symmetric contour features to strengthen the perception of target geometric structures; a cascaded multi-receptive field module is embedded at the end of the trunk to integrate multi-scale features in a hierarchical manner to take into account both expressive ability and computational efficiency with a focus on fusing symmetric multi-scale features to optimize feature representation; the neck is integrated with a spatially adaptive feature modulation network to achieve dynamic weighting of cross-layer features and detail fidelity and, meanwhile, models symmetric feature dependencies across channels to reduce the loss of discriminative information. Experimental results based on the VisDrone2019 data set show that ACS-YOLOv8s is superior to the baseline model in precision, recall, and mAP indicators, with mAP50 increased by 2.8% to 41.6% and mAP50:90 increased by 1.9% to 25.0%, verifying its effectiveness and robustness in small-target detection in complex drone aerial-photography scenarios. Full article

Automated assessment of coronary artery (CA) lesions via Coronary Computed Tomography Angiography (CCTA) is essential for the diagnosis of coronary artery disease (CAD). However, current deep learning approaches confront several challenges, primarily regarding the modeling of long-range anatomical dependencies, the effective decoupling of plaque texture from stenosis geometry, and the utilization of clinically prevalent mixed-grained annotations. To address these challenges, we propose a novel mixed-grained hierarchical geometric-semantic learning network (MG-HGLNet). Specifically, we introduce a topology-aware dual-stream encoding (TDE) module, which incorporates a bidirectional vessel Mamba (BiV-Mamba) encoder to capture global hemodynamic contexts and rectify spatial distortions inherent in curved planar reformation (CPR). Furthermore, a synergistic spectral–morphological decoupling (SSD) module is designed to disentangle task-specific features; it utilizes frequency-domain analysis to extract plaque spectral fingerprints while employing a texture-guided deformable attention mechanism to refine luminal boundary. To mitigate the scarcity of fine-grained labels, we implement a mixed-grained supervision optimization (MSO) strategy, utilizing anatomy-aware dynamic prototypes and logical consistency constraints to effectively leverage coarse branch-level labels. Extensive experiments on an in-house dataset demonstrate that MG-HGLNet achieves a stenosis grading accuracy of 92.4% and a plaque classification accuracy of 91.5%. The results suggest that our framework not only outperforms state-of-the-art methods but also maintains robust performance under weakly supervised settings, offering a promising solution for label-efficient CAD diagnosis. Full article

Underwater image enhancement remains challenging due to wavelength-dependent absorption, spatially varying scattering, and non-uniform illumination, which jointly cause severe color distortion, contrast degradation, and structural information loss. To address these issues, we propose UCS-Net, a degradation-aware dual-path framework that exploits the complementarity between global and local representations. A spatial color balance module first stabilizes the chromatic distribution of degraded inputs through a learnable gray-world-guided normalization, mitigating wavelength-induced color bias prior to feature extraction. The network then adopts a dual-branch architecture, where a hierarchical Swin Transformer branch models long-range contextual dependencies and global color relationships, while a multi-scale residual convolutional branch focuses on recovering local textures and structural details suppressed by scattering. Furthermore, a multi-scale attention fusion mechanism adaptively integrates features from both branches in a degradation-aware manner, enabling dynamic emphasis on global or local cues according to regional attenuation severity. A hue-preserving reconstruction module is finally employed to suppress color artifacts and ensure faithful color rendition. Extensive experiments on UIEB, EUVP, and UFO benchmarks demonstrate that UCS-Net consistently outperforms state-of-the-art methods in both full-reference and non-reference evaluations. Qualitative results further confirm its effectiveness in restoring fine structural details while maintaining globally consistent and visually realistic colors across diverse underwater scenes. Full article

Drug–target affinity (DTA) prediction is one of the core components of drug discovery. Despite considerable advances in previous research, DTA tasks still face several limitations with insufficient multi-modal information of drugs, the inherent sequence length limitation of protein language models, and single attention mechanisms that fail to capture critical multi-scale features. To alleviate the above limitations, we developed a multi-granularity fusion model for drug–target binding affinity prediction, termed MGF-DTA. This model is composed of three fusion modules, specifically as follows. First, the model extracts deep semantic features of SMILES strings through ChemBERTa-2 and integrates them with molecular fingerprints by using gated fusion to enhance the multi-modal information of drugs. In addition, it employs a residual fusion mechanism to integrate the global embeddings from ESM-2 with the local features obtained by the k-mer and principal component analysis (PCA) method. Finally, a hierarchical attention mechanism is employed to extract multi-granularity features from both drug SMILES strings and protein sequences. Comparative analysis with other mainstream methods on the Davis, KIBA, and BindingDB datasets reveals that the MGF-DTA model exhibits outstanding performance advantages. Further, ablation studies confirm the effectiveness of the model components and case study illustrates its robust generalization capability. Full article

In recent years, extracting coastlines from high-resolution remote sensing imagery has proven difficult due to complex details and variable targets. Current methods struggle with the fact that CNNs cannot model long-range dependencies, while Transformers incur high computational costs. To address these issues, we propose CVMFusion: a land–sea segmentation network based on a U-shaped encoder–decoder structure, whereby both the encoder and decoder are hierarchically organized. This architecture integrates the local feature extraction capabilities of CNNs with the global interaction efficiency of Mamba. The encoder uses parallel ConvNeXtV2 and VMamba branches to capture fine-grained details and long-range context, respectively. This network incorporates Dynamic Multi-Scale Attention (DyMSA) and Dynamic Weighted Cross-Attention (DyWCA) modules, which replace the traditional concatenation with an adaptive fusion mechanism to effectively fuse the features from the dual-branch encoder and utilize skip connections to complete the fusion between the encoder and decoder. Experiments on two public datasets demonstrate that CVMFusion attained MIoU accuracies of 98.05% and 96.28%, outperforming existing methods. It performs particularly well in segmenting small objects and intricate boundary regions. Full article

Background/Objectives: Cognitive decline and compromised attention control serve as early indicators of neurodysfunction that manifest as broader psychopathological symptoms, yet conventional mental health assessment relies predominantly on subjective self-report measures lacking objectivity and temporal granularity. We propose Neurosense, an AI-driven brain health assessment framework using electroencephalography (EEG) to non-invasively capture neural dynamics. Methods: Our Dual-path Spatio-Temporal Adaptive Gated Encoder (D-STAGE) architecture processes temporal and spatial EEG features in parallel through Transformer-based and graph convolutional pathways, integrating them via adaptive gating mechanisms. We introduce a two-stage paradigm: first training on cognitive task EEG for reaction time prediction to acquire cognitive performance-related representations, then featuring parameter-efficient adapter-based transfer learning to estimate p-factor—a transdiagnostic psychopathology dimension. The adapter-based transfer achieves competitive performance using only 1.7% of parameters required for full fine-tuning. Results: The model achieves effective reaction time prediction from EEG signals. Transfer learning from cognitive tasks to mental health assessment demonstrates that cognitive efficiency representations can be adapted for p-factor prediction, outperforming direct training approaches while maintaining parameter efficiency. Conclusions: The Neurosense framework reveals hierarchical relationships between neural dynamics, cognitive efficiency, and mental health dimensions, establishing foundations for a promising computational framework for mental health assessment applications. Full article

Most Transformer-based tracking methods overemphasize tracking accuracy while neglecting tracking efficiency. To address this issue, a lightweight object tracking method based on a hierarchical pruning strategy is proposed. First, an activation module is introduced to adaptively adjust the attention layers of the backbone network, avoiding unnecessary computations of attention layers. Then, based on the demands at different stages of the network, the standard attention mechanism is improved and divided into hybrid attention and cross attention, further reducing redundant computations. Experimental results on multiple benchmark datasets demonstrate that PSTrack achieves a tracking speed of 216 FPS on GPU and 48 FPS on CPU while maintaining competitive accuracy, with an AO of 68.7% on GOT-10k and an AUC of 66.1% on LaSOT. The proposed tracking method exhibits strong competitiveness in both quantitative and qualitative evaluations. Full article

Rapid and accurate acquisition of landslide inventories is essential for effective disaster relief. Deep learning-based pixel-wise semantic segmentation of remote sensing imagery has greatly advanced in landslide mapping. However, the heavy dependance on extensive annotated labels and sensitivity to domain shifts severely constrain the model performance in unseen domains, leading to poor generalization. To address these limitations, we propose LandsDANet, an innovative unsupervised domain adaptation framework for cross-domain landslide identification. Firstly, adversarial learning is employed to reduce the data distribution discrepancies between the source and target domains, thereby achieving output space alignment. The improved SegFormer serves as the segmentation network, incorporating hierarchical Transformer blocks and an attention mechanism to enhance feature representation capabilities. Secondly, to alleviate inter-domain radiometric discrepancies and attain image-level alignment, a Wallis filter is utilized to perform image style transformation. Considering the class imbalance present in the landslide dataset, a Rare Class Sampling strategy is introduced to mitigate bias towards common classes and strengthen the learning of the rare landslide class. Finally, a contrastive loss is adopted to further optimize and enhance the model’s ability to delineate fine-grained class boundaries. The proposed model is validated on the Potsdam and Vaihingen benchmark datasets, followed by validation in two landslide scenarios induced by earthquakes and rainfall to evaluate its adaptability across different disaster domains. Compared to the source-only model, LandsDANet achieved improvements in IoU of 27.04% and 35.73% in two cross-domain landslide disaster recognition tasks, respectively. This performance not only showcases its outstanding capabilities but also underscores its robust potential to meet the demands for rapid response. Full article

Memory resident malware, particularly fileless and heavily obfuscated types, continues to pose a major problem for endpoint defense tools, as these threats often slip past traditional signature-based detection techniques. Deep learning has shown promise in identifying such malicious activity, but its use in real Security Operations Centers (SOCs) is still limited because the internal reasoning of these neural network models is difficult to interpret or verify. In response to this challenge, we present HAGEN, a hierarchical attention architecture designed to combine strong classification performance with explanations that security analysts can understand and trust. HAGEN processes memory artifacts through a series of attention layers that highlight important behavioral cues at different scales, while a gated mechanism controls how information flows through the network. This structure enables the system to expose the basis of its decisions rather than simply output a label. To further support transparency, the final classification step is guided by representative prototypes, allowing predictions to be related back to concrete examples learned during training. When evaluated on the CIC-MalMem-2022 dataset, HAGEN achieved 99.99% accuracy in distinguishing benign programs from major malware classes such as spyware, ransomware, and trojans, all with modest computational requirements suitable for live environments. Beyond accuracy, HAGEN produces clear visual and numeric explanations—such as attention maps and prototype distances—that help investigators understand which memory patterns contributed to each decision, making it a practical tool for both detection and forensic analysis. Full article